Sequential Estimates of the Parameters in a Random Coefficient Autoregressive Process
D. V. Kashkovsky and V. V. Konev
Tomsk State University, Tomsk, Russia E-mail: vvkonev@mail.tsu.ru
Pages: 52-61
Abstract
A one-step sequential procedure using a special rule of observation accumulation is proposed for estimating the linear parameters of a stable random coefficient autoregressive process. The upper bound for rms precision of a sequential estimate which decreases inversely to the value of the procedure parameter that determines the time of observations is obtained. Asymptotic behavior of the mean time of the procedure is studied. Results of numerical simulations are presented.
|
